Cognitive behavioural computer therapy for the anxiety disorders: A ...

6 downloads 19211 Views 138KB Size Report
Address for Correspondence: Dr Jim White, Consultant Clinical Psychologist, The ... interactive computer package capable of providing a rapid, effective and ...
Journal of Mental Health (2000) 9, 5, 505 –516

Cognitive behavioural computer therapy for the anxiety disorders: A pilot study JIM WHITE1, RAY JONES2 & EILEDH MCGARRY2 Greater Glasgow Community and Mental Health Services NHS Trust & 2Department of Public Health, University of Glasgow, Glasgow, Scotland 1

Abstract As part of a wider attempt to improve treatment provision in primary care and to provide greater patient choice, a pilot study of a three session CD-ROM anxiety management programme was tested on 26 carefully assessed and representative long-term sufferers of anxiety disorder living in an area of high social deprivation. The aim was to test whether a successful written package could be turned into an interactive computer package capable of providing a rapid, effective and inexpensive intervention to those patients who wish such an approach. Treatment outcome to 6-month follow-up suggests that this approach was effective and was welcomed by patients.

Introduction The US National Co-morbidity Survey (Kessler, 1997) suggested that 48% of the population would meet criteria for a mental health disorder at some stage of their lives. Seventeen per cent of this large community sample met criteria for an anxiety disorder in the previous 12 months, making anxiety disorders the most common mental health problem. At the primary care level one quarter to one third of all GP consultations are for mental health problems (Sharp & Morell, 1989; Shah, 1992). Approximately half of these problems will be missed or misdiagnosed by the GP (Fifer et al., 194). Routinely screening primary care patients using the Hospital Anxiety and Depression scale (Snaith & Zigmond, 1994) resulted in 51% being ‘probable cases’ and 28% as

‘cases’ of psychiatric disorder (Dowell & Biran, 1990 ); anxiety and depression made up the vast majority of these problems (Lloyd & Jenkins, 1995). The British Psychiatric Morbidity Survey (Jenkins et al., 1998) reported that one in six adults had suffered from some type of neurotic disorder in the week prior to interview. Lloyd et al. (1996) report that almost half of a primary care patient sample followed over 11 years had a chronic course with over half this number meeting ‘caseness’ at the 11 year assessment point. They also noted a significantly higher mortality rate (from natural causes) than would be expected. GPs are highly likely to be the sole health care provider to the vast majority of these patients (Espie & White, 1986); secondary care services see only the tip of the iceberg. Clinical Psychologists have, on average,

Address for Correspondence: Dr Jim White, Consultant Clinical Psychologist, The Greater Glasgow Primary Care NHS Trust, Clydebank Health Centre, Kilbowie Road, Clydebank G81 2TQ, Scotland. Tel: 0141 531 6300 Fax: 0141 531 6336, E-mail: [email protected] ISSN 0963-8237print/ISSN 1360-0567online/2000/050505-12 © Shadowfax Publishing and Taylor & Francis Ltd DOI: 10.1080/09638230020005237

506

Jim White et al.

waiting lists of 17.5 weeks (Division of Clinical Psychology, 1997) while controlled evidence is beginning to suggest that personcentred counselling, now available to over half of general practices in England and Wales, is no better than routine GP care in dealing with heterogeneous mental health problems (Friedli et al., 1997; Harvey et al., 1998). Similar results were obtained for CPNs (Gourney & Brooking, 1994). It may be that future studies, looking at more specific disorders within primary care, may yield different results. However, it does seem clear that there is a need to radically rethink our approach to these common disorders. In addition, patients and GPs have only rarely been offered a range of services that might better meet the heterogeneous problems presenting in primary care. Service provision might be dependent on the particular interests and skills of the therapist(s) covering a particular locality with one-to-one therapy predominating. To better meet the needs of the large number of people who could potentially benefit from therapy, therapists must improve access to services, offer wider choice and intervene sooner. On the latter point, data from the US National Comorbidity Study suggested that there was a strong inverse effect of age at onset and first treatment contact both in the US and in Ontario (Olfson et al., 1998). As Kessler and colleagues have pointed out in a series of papers, this often resulted in detrimental social consequences – failure to complete high school, teenage parenthood, increased likelihood of divorce, lower socio-economic status (e.g. Kessler & Forthofer, 1999). In order to alleviate the difficulties in reaching appropriate help and as we cannot expect to significantly increase staff numbers, we must look to alternative approaches emphasising less time intensive interventions. Cognitive –behavioural therapies (CBT) have been shown

to be reasonably effective both in therapeutic and time-efficiency terms (Chambliss & Gillis, 1993 ). Although often tested on almost perfect patients (i.e. those who meet such strict inclusion criteria that they are unrepresentative of typical patients seen in clinics) and treated in almost perfect circumstances (i.e. by clinicians who do not carry heavy caseloads and who have time to follow detailed treatment protocols), it does appear to be malleable enough to meet the demands of routine clinical settings (e.g. White et al., 1992). Personalised computer-based therapies offer the chance to augment our existing repertoire of interventions, although as Kenardy & Adams (1993 ) note, the potential advantages for the use of computers seem to have been overlooked in clinical practice. Cognitive – behavioural therapy seems particularly well suited to computerisation due to its well delineated and systematically implemented procedures. With ever increasing numbers of people using computers in their daily lives and with the power and multimedia options of computers increasing markedly year by year while, at the same time, costs decrease, computer based therapies would appear to offer potential benefits to over-worked and under-resourced clinicians. Computer based interviews have been shown to be acceptable to patients (e.g. Erdman et al., 1992) and even preferable to contact with a therapist (e.g. Petrie & Abell, 1994). Patients may reveal more information to a computer than to a therapist (Fowler, 1985) and feel less embarrassment disclosing sensitive personal information (e.g. Kobak et al., 1994). In an excellent review of the area, Newman et al. (1997b) provide evidence, comparing computer- and standard- therapy, showing equivalent drop-out rates (Ghosh & Marks, 1987) and equivalent rates of satisfaction (Ghosh et al., 1988). Compliance in this agoraphobia

Computer therapy 507

study was higher for the computer therapy group. In terms of therapeutic outcome, there is evidence to suggest that computer therapy is an acceptable approach in the treatment of anxiety (Carr et al., 1988). While many studies suffer from various methodological weaknesses – such as small sample sizes, non-representative samples and lack of control groups – Newman et al. (1997a) have produced controlled evidence for the effectiveness of palm-top computer treatment for panic disorder. Compared to 12-session CBT, the computerised therapy, involving four therapist contact sessions and use of a palmtop computer for 12 weeks, produced, on some measures, less clinically significant change at post-therapy. This difference disappeared at 6-month follow-up when 46% of the standard therapy and 35% of the computer therapy patients achieved clinically significant improvement. Thus there is some evidence that cognitive –behavioural therapy can be successfully transferred to this modality. White (1995; 1998a) has shown, in a controlled study, that a self-help written CBT package for the anxiety disorders (‘Stresspac’, White, 1997) produced clinically significant improvement in 67% of the sample at posttherapy, 89% at 1-year follow-up and 78% at 3-year follow-up. As with a study looking at large group didactic therapy for Generalised Anxiety Disorder – ‘Stress Control’ (White, 1998b), the most highly rated components of therapy were the provision of readily understood, personally relevant information about anxiety. It was suggested that one of the main factors in explaining change was that the package allowed patients to perceive that they had the ability to regain control over areas of their lives in which they had previously felt control to be lacking. This effect seemed independent of therapist-contact thus

possibly explaining why maintenance or further improvement was found at long term follow-up. Computerised therapy, again, may offer the same therapeutic factors while offering the prospect of greater personalisation of therapy. This pilot study looked to develop and test a computerised therapy for anxiety with minimal therapist contact and robust enough to deal with the heterogeneous problems typically found in patients presenting to primary care services (see Parkin et al., 1995). In particular, we asked the following questions: • Can an existing written self-help approach be transferred into a computerised modality? • Is the system acceptable to representative patients from a socially deprived area? • Do they see the package as personally relevant, easily understandable and are the treatment sections seen as appropriate? • Does a short intervention (maximum three sessions) result in symptomatic improvement and, if so, is the therapy robust enough to allow maintenance of any treatment gains at 6 months follow-up without further therapist contact?

Method Computer system The system was developed using Adobe Premier to digitise video and Macromedia Authorware for Windows for system development (and possible future delivery of the package on the WWW). An approach was devised incorporating text, graphics, voiceover, video, animation and music. A 17-inch touch screen was used in preference to mouse and keyboard to make the system easier to use and less frightening to non-computerliterate patients. Personalisation was achieved

508

Jim White et al.

by using the Beck Depression Inventory (Beck & Steer, 1987) and Beck Anxiety Inventory – BAI (Beck & Steer, 1987). The latter allowed the extraction of a cognitive and somatic score. Along with additional on-screen questions on panic, insomnia and behavioural anxiety, the computer stored these data and, in sessions 2 and 3, offered a tailored treatment to each individual. We aimed at as straight forward an approach as possible. This was partly due to the nature of anxiety where concentration problems and self-doubt about mastering a novel task could interfere with treatment compliance and, in the fourth most socially deprived district in Scotland, we felt it crucial, due to widespread literacy problems, to limit the amount of text on-screen. A voice-over (JW) imparted more detailed information. Pragmatically, three sessions seemed more reasonable in terms of encouraging GPs to recommend the approach to patients. The contents of each session were as follows: Session 1 Patients were logged on using their name and date of birth. The session began with a video clip of James McPherson (star of British television’ s ‘Taggart’) introducing the system to the patients. In this McPherson talked about stress, how it affected him while filming and how the computer treatment could help. The patient was given a brief introduction to the system, during which they practised using the touch screen, familiarised themselves with the buttons that appeared on the screen, and were given an overview of the session. Patients then completed, on-screen, the Beck Anxiety Inventory (Beck & Steer, 1990 ) and Beck Depression Inventory (Beck and Steer, 1987). The computer fed back the individual’ s score in terms of a bar chart immediately and explained the significance of the score. The system also informed the research assistant if the individual reported

suicidal ideation. The session continued with descriptions of what anxiety is and how it affects cognitions, behaviour and somatics (‘thoughts, actions and body’ ). A log file was created for the patient during the session that included information, such as patients’ responses to questions asked on-screen and duration of programme sections. Sessions were expected to last around 40 minutes. At the end of each session, a print-out of BAI and BDI scores were given. Session 2 – treatment The session started with a reminder of the buttons used in the system and an overview of the session. Patients completed the on-line BAI and BDI and were given their results along with the scores from the previous session. Patients were then presented with a menu containing six colour-coded treatment options: Learning relaxation Controlling panic attacks Controlling stressful thoughts Getting a good night’ s sleep Facing up to stress Coping with the future Based on their on-line assessment, the computer recommended which options were most relevant. Patients were free to ignore this and could visit each option twice. After the patient left each option, it turned to grey on screen. Patients could then open another option or end the session. At the end of the session a printout of the options viewed was given. Session 3 – treatment Similar to session 2. The aim was for patients to view options not yet seen or to revisit options seen in session 2. On-line BAI and BDI were again completed and the scores printed. This session was optional as patients may have seen all they required in session 2.

Computer therapy 509

Handouts Due to concentration problems in anxiety, eight written handouts were also made available – six related to the treatment options. These varied between 2 and 8 pages. An ‘Introduction to the treatment’ (two pages) set expectations, while an 18 page ‘Information’ handout summarised the material in session 1. Using Flesch reading ease scores (Flesch, 1948), all handouts should be readily understood by 86% of the population (IQ of 84+, Ley, 1977). Diary forms, for instance, for use in assessing panic attacks and controlling thoughts, were offered at the appropriate point along with a relaxation tape for patients viewing the ‘Learning relaxation’ option. The handouts were designed to act as retrieval cues for the material presented and to help prevent relapse. Procedure GPs in a large Health Centre were informed of the project and asked to refer mild-tomoderately anxious individuals who would not normally be referred to the clinical psychology primary care service based in the Centre. These individuals were then informed of the project and invited for assessment by JW, using the Anxiety Disorder Interview Schedule-IV (Brown et al., 1994) allowing DSM-IV diagnoses (APA, 1994). Exclusion criteria were kept to a minimum to ensure a representative population. These exclusion criteria included evidence of psychosis, marked suicidal ideation, severe drug or alcohol abuse, recent change in psychotropic medication, concurrent secondary-care mental health treatment or cognitive –behavioural treatment within 3 years. For pragmatic and theoretical reasons (e.g. Brown et al., 1998), patients were included if they met criteria for a principal diagnosis of any DSMIV anxiety disorder and scored at least eight on the anxiety scale of the Hospital Anxiety

and Depression Scale (Snaith & Zigmond, 1994). Zinbarg & Barlow (1996) have presented a hierarchical model of anxiety that suggests that, while the various anxiety disorders can be differentiated at one structural level, a higher order factor – ‘negative affectivity’ – represents a trait diathesis common to all these disorders. This suggests that, along with good therapeutic reasons (e.g. White, 1995; Tyrer et al., 1993 ) we should look to develop approaches tackling the commonalities shared by these disorders. In addition, such pragmatic approaches are more likely to be picked up and used by mental health workers working in busy clinical settings. Patients meeting criteria were contacted by the research assistant – EMcG (a science graduate with no psychological experience) who remained in the room with the individual but did not offer any help unless asked. Two sessions were offered within 1 week and the third 2 weeks later. Following session 3, no further help was offered. Patients Of 38 patients referred, 33 met criteria for the study. All accepted. Two were unable to attend during the day and alternative treatment was given. Of the remaining 31, all attended at least one session. However, one was unable to return as the project had to end and two did not agree that they had an anxiety problem. It is not known why the remaining two patients failed to attend. These five patients are excluded from analyses. Thus, the sample comprises 26 patients who judged themselves to have completed treatment, i.e. attended at least two sessions. Patient characteristics Nineteen females and seven males, mean age 35 (range 17–67) took part. Of these, 84% had received no further education following secondary school; 45% had had previ-

510

Jim White et al.

ous treatment with either a psychiatrist (30%) or clinical psychologist (15%). Average duration of current episode symptoms was 4.6 years. Half were currently using anti-depressant medication, eight per cent beta-blockers and 12% benzodiazepines; all had been taking them for more than 4 months. ADIS-IV assessment showed that, of the 26, seven (27%) had a principal diagnosis of panic disorder with agoraphobia, two (8%) met criteria for panic disorder without agoraphobia, 12 (46%) generalised anxiety disorder, four (15%) had social phobia and one (4%) anxiety disorder not otherwise specif ied. Also, 15% did not meet criteria for any other Axis-I disorder, 15% met criteria for one, 47% for two and 23% for three additional diagnoses. Thirty-eight per cent of all patients also met criteria for major depression and 23% for dysthymia. Relating to computer literacy, 62% had either used computers at school or at work. None had access to a home computer. Measures Patients completed the Hospital Anxiety and Depression Scale – HADs (Snaith & Zigmond, 1994) and the Brief Symptom Inventory (Derogatis, 1993 ). Two scores were yielded from this – the General Severity Index (GSI) and the Positive Symptom Total (PST). Higher scores in all measures represent greater distress. These measures were administered by post prior to assessment interview, at post-therapy and at 6-month follow-up. On-line BAI and BDIs were completed in each session. In addition, credibility, satisfaction and other ratings were taken throughout.

Results Just over half (54%) of patients required three sessions, the remaining completed all the options they wished to see in two ses-

sions. The average length of sessions 1 and 2 was 40 minutes and 32 minutes for session 3. There was a non-significant trend for older patients to take more time. Credibility and other ratings All patients interviewed at the end of session 1 reported they had ‘got on well’. Although 38% had no prior experience of computers, no one reported any difficulty in using the package. About half (48%) of those who were experienced computer users preferred to use the mouse rather than the touch-screen. At the end of session 2, all users reported that the computer understood them and described problems in a way they understood and 27% wanted greater personalisation, e.g. data from medical records, to be included. Of the 14 patients who attended for the third session, all felt it to be a useful source of information and significantly decreased their sense of isolation. All felt this way of learning about how to cope with stress was a good idea both in general and for them personally. Asked to compare the treatment with a popular anxiety management evening class run in the centre ‘Stress Control’ (White, 1998b), 44% said they would prefer the computer, 44% would accept either and the remaining 12% preferred the idea of the class. Advantages for the computer package were seen as no waiting time, flexible appointments, privacy and the ability to work at one’s own pace. Overall, 83% would recommend the approach to relatives or friends. Similarly high rates of patients reported that they had increased their understanding of stress. All patients reported they had read and benefited from the handouts. Half had changed their view on what had caused their stress following the computer therapy. Outcome measures Means and standard deviations for HADs and BSI measures are shown in Table 1 and

Computer therapy 511 Table 1: Mean (and standard deviation) scores of HAD: Anxiety, HAD: Depression, BSI:GSI and BSI:PST at pre-therapy, post-therapy and 6-month follow-up HADAnxiety

HADDepression

BSI-GSI

BSI-PST

Pre-therapy (n=26)

15.4 (2.19)

11.51 (3.24)

2.05 (.49)

45.9 (10.02)

Post-therapy (n=25)

12.56 (2.8)

9 (3.7)

1.57 (.47)

36.9 (11.5)

6 month follow-up (n=21)

9.19 (2.8)

6.76 (2.8)

1.24 (.48)

31.8 (8.7)

HAD-anxiety

16 n

15

HAD-depression

12 u

11

14

10

13

9

12 11

8

n

10

u

7

9

6

8 n

7

Pretherapy

6-month follow-up

BSI-GSI

2.2 2

Posttherapy

u

5 Pretherapy

Posttherapy

6-month follow-up

BSI-PST

50

s

45

t

1.8 40

1.6 1.4

35

s

1.2

t

30

1

t

s

0.8 Pretherapy

Posttherapy

6-month follow-up

25

Pretherapy

Posttherapy

6-month follow-up

Figures 1–4: HAD-anxiety and depression, BSI–GSI and PST pre- and post-therapy and 6-month follow-up.

Figures 1– 4. One-way repeated measure MANOVAS showed significant change on HAD: Anxiety, HAD: Depression, BSI-GSI and BSI-PST. Newman–Keuls tests showed sig-

nificant differences on all measures between pre-therapy and follow-up and between posttherapy and follow-up (p